In the early days of computers they used to use a length of wire as
temporary memory. At the start of a store cycle a piece of data would be
input to the wire, after a period of time the data would come out and be
placed into the computation.
Admiral Grace Hopper used to give an example of time and delay in her
speeches. She would say that one day she called down to the computer
department and asked for a micro second. They sent her 1000 feet of
wire. She then called down and asked for a nanosecond, they sent her one
foot of wire. No point to this just a good story.

Dave N
The delay lines I encountered at IBM that were used for storage were
"sonic" delay lines. They were driven in a torsion mode (mechanical)
and were very reliable up to about 10 milliseconds(I think!) of data.
Above that they began to be temperamental and required constant
temperature ovens. The larger ones were used as video storage with a
screen of data in each instance.

Delay lines running at light speed (about a NS per ft) were used as
clock generators by inputting a pulse and tapping the line down stream
for a very stable clock sequence.

John Ferrell W8CCW

A feature of the '50s-era radars I worked on in the '60s was MTI, or
moving target indication. It was done simply by subtracting the return
from the previous pulse from the current one and displaying only things
which had changed. This was done to reduce clutter from fixed objects,
and it was effective against some jamming techniques. Doing something
like that is trivial today, but it wasn't back then. The pulse
repetition time of the long-range radars was over 2 ms (round trip time
for 400 miles or so), so a delay of that time with a bandwidth of a few
MHz was required.

The sets I worked on used a large piezoelectric quartz slab with many
faces, and transducers on two of the faces. One transducer would convert
the electrical signal to a mechanical wave which would enter the slab,
bounce around from one face to another, until at the right time it would
hit a face below the critical angle and exit. And that face was, of
course, where the other transducer was. To keep the timing precise, the
slab was used to control the radar pulse interval.

Not long before I was involved, a mercury delay line was used. This was
a tube of mercury with a transducer at each end, and a mechanical wave
was sent from one end to the other. I never saw one, but heard many
stories about how fussy they were and that it would take hours or days
for waves to settle if it was bumped or jiggled. Folks who had dealt
with them considered the quartz delay line to be a big improvement.

Roy Lewallen, W7EL

Roy Lewallen wrote in news:12tkkh6n7us4v19
@corp.supernews.com:

A feature of the '50s-era radars I worked on in the '60s was MTI, or
moving target indication. It was done simply by subtracting the return
from the previous pulse from the current one and displaying only things
which had changed. This was done to reduce clutter from fixed objects,
and it was effective against some jamming techniques. Doing something
like that is trivial today, but it wasn't back then. The pulse
repetition time of the long-range radars was over 2 ms (round trip time
for 400 miles or so), so a delay of that time with a bandwidth of a few
MHz was required.


My recollection was that LC delay lines were also used, and MTI radar was
very useful for close in to the airport where structures (the built
environment) created the greatest clutter problem distinguishing low
flying aircraft, so shorter delay times (hundreds of us) were adequate.

Other techniques I have come across for various delay applications,
including circulating memories included various forms of semi-distributed
LC elements, coils of piano wire with magnetostrictive transducers, and
early dynamic shift registers that were a kind of charge shuffling
arrangement.

Owen